As used herein, the term "ligand-receptor assay" refers to an assay for at least one target ligand which may be detected by the formation of a complex between the ligand and a receptor capable of specific interaction with that target ligand. The target ligand may be the analyte itself or a substance which, if detected, can be used to infer the presence of the analyte in a sample. In the context of the present invention, the term "ligand", includes haptens, hormones, peptides, proteins, deoxyribonucleic acid (DNA), ribonucleic acids (RNA), metabolites of the aforementioned materials and other substances of either natural or synthetic origin which may be of diagnostic interest and have a specific ligand receptor therefor. Ligand-receptor assays are generally useful for the in vitro determination of the presence and concentration of ligands in body fluids, food products, animal fluids, and environmental samples. For example, the determination of specific hormones, peptides, proteins, therapeutic drugs, and toxic drugs in human blood or urine has significantly improved the medical diagnosis of the human condition. There is a continuing need for improvements in such assays in order to increase their accuracy and reliability.
Ligand-receptor assays rely on the binding of target ligands by ligand receptors to determine the concentrations of target ligands in a sample. Ligand-receptor assays can be described as either competitive or non-competitive. Competitive assays generally involve a sample suspected of containing target ligand, a ligand analogue conjugate, and the competition of these species for a limited number of binding sites provided by the ligand receptor.
Non-competitive assays generally utilize ligand receptors in substantial excess over the concentration of target ligand to be determined in the assay. Sandwich assays, in which the target ligand is detected by binding to two ligand receptors, one ligand receptor labeled to permit detection and a second ligand receptor, frequently bound to a solid phase, to facilitate separation from unbound reagents, such as unbound labeled first ligand receptor, are examples of non-competitive assays. Methods utilizing two monoclonal antibodies, selected to bind the antigenic substance (target ligand) at sites remote from each other so as to not interfere with the others binding to the antigen, in sandwich assays are described in U.S. Pat. No. 4,376,110. Similar assays for the determination of haptens are described in International Application Number PCT/US84/01737. While such assays are designed so that the concentration of the receptors is each in excess over the concentration of the target ligand in the assay range, some target ligands can be present in samples at concentrations that are substantially higher than the concentrations of receptors employed in the assay. In simultaneous, sandwich assays where the labeled receptor and the unlabeled receptor are mixed together with the sample, a large excess of the target ligand can result in binding of separate target ligand molecules to the labeled receptor and to the unlabeled receptor so that the formation of the sandwich complex of labeled receptor/target ligand/unlabeled receptor is inhibited. The response in such assays can be misinterpreted so that the determination of the target ligand concentration can result in an incorrect concentration, a much lower concentration than is actually present in the sample. This is widely known as the "hook" effect in sandwich assays. Faced with such a possibility, users of such assays routinely assay dilutions of the sample to determine if the dilutions are quantitated linearly (i.e. the concentration of the target ligand determined for the dilution, when multiplied by the dilution factor, is the same as the target ligand concentration determined for the sample). Such additional testing would be unnecessary if the "hook" effect were not a potential problem. The hook effect can be minimized in sandwich assays by choosing sequential assay protocols where the unlabeled receptor is immobilized on a solid phase and the solid phase is washed to remove unbound target ligand after incubation with the sample and before addition of the labeled receptor. Such assay protocols are lengthy and require more steps and manipulation than simultaneous protocols.
The problem of "hook" effects in sandwich assays has been addressed in several ways. In U.S. Pat. No. 4,743,542, a method is described where one of the receptors is labeled for detection and the other receptor is labeled with a hapten so that a receptor for the hapten can be used to bind the hapten-labeled receptor to a solid phase. The invention utilizes either unlabeled first receptor or non-haptenated second receptor to extend the assay range for the target ligand by minimizing the "hook" effect. The requirement for additional receptor is a principal disadvantage of this method because in some cases the concentration of target ligand can be so high that the quantity of additional receptor that is needed to prevent the "hook" effect is not practical. Similarly, the method of U.S. Pat. No. 4,778,751 requires excess receptor coupled to a liquid matrix that can be immobilized on a solid phase. Again, the primary mechanism used to overcome the "hook" effect is the use of large quantities of receptor needed to bind up all of the target ligand, an impractical solution for many target ligands.
In the present invention antibodies are selected that bind the complex of ligand receptor and target ligand (ligand analogue conjugate) and substantially do not bind the ligand receptor or the target ligand when they are not bound to one another. The use of antibodies to ligand receptor-ligand complexes in sandwich assays eliminates or greatly reduces the "hook" effect without the requirement for excessive amounts of such antibodies.
Antibodies that bind the complex of ligand and a specific antibody for the ligand have been described by Nemazee and Sato (Proc. Natl. Acad. Sci. USA, Vol. 79, pp. 3828-3832, 1982). They postulated that three types of antibody are produced in response to immunization with antibody-antigen complexes, antibodies that bind to either conformationally altered antibody or antigen and antibodies that bind to parts of both antibody and antigen. Nemazee and Sato also provide methods for producing antibodies of the first type, antibodies that bind conformationally altered antibody when it is bound to antigen. Johannsson, in UK Patent Application No. 8505487, describes antibodies produced by immunization of the complex of a fragment of specific antibody and its ligand. The resulting antibodies are said to bind the complex of a specific binding fragment and its binding partner with high affinity while binding the specific binding fragment or the binding partner with low affinity. This prior art does not describe methods for selecting and using such antibodies to overcome the "hook" effect in sandwich assays. In the present invention the use of antibodies to complexes of ligands and ligand receptors is described in embodiments that substantially eliminate the "hook" effect in sandwich ligand-receptor assays.